yt_dlp/extractor/vine.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4
   5 from .common import InfoExtractor
   6 from ..compat import compat_str
   7 from ..utils import (
   8     determine_ext,
   9     format_field,
  10     int_or_none,
  11     unified_timestamp,
  12 )
  13
  14
  15 class VineIE(InfoExtractor):
  16     _VALID_URL = r'https?://(?:www\.)?vine\.co/(?:v|oembed)/(?P<id>\w+)'
  17     _TESTS = [{
  18         'url': 'https://vine.co/v/b9KOOWX7HUx',
  19         'md5': '2f36fed6235b16da96ce9b4dc890940d',
  20         'info_dict': {
  21             'id': 'b9KOOWX7HUx',
  22             'ext': 'mp4',
  23             'title': 'Chicken.',
  24             'alt_title': 'Vine by Jack',
  25             'timestamp': 1368997951,
  26             'upload_date': '20130519',
  27             'uploader': 'Jack',
  28             'uploader_id': '76',
  29             'view_count': int,
  30             'like_count': int,
  31             'comment_count': int,
  32             'repost_count': int,
  33         },
  34     }, {
  35         'url': 'https://vine.co/v/e192BnZnZ9V',
  36         'info_dict': {
  37             'id': 'e192BnZnZ9V',
  38             'ext': 'mp4',
  39             'title': 'ยิ้ม~ เขิน~ อาย~ น่าร้ากอ้ะ >//< @n_whitewo @orlameena #lovesicktheseries  #lovesickseason2',
  40             'alt_title': 'Vine by Pimry_zaa',
  41             'timestamp': 1436057405,
  42             'upload_date': '20150705',
  43             'uploader': 'Pimry_zaa',
  44             'uploader_id': '1135760698325307392',
  45             'view_count': int,
  46             'like_count': int,
  47             'comment_count': int,
  48             'repost_count': int,
  49         },
  50         'params': {
  51             'skip_download': True,
  52         },
  53     }, {
  54         'url': 'https://vine.co/v/MYxVapFvz2z',
  55         'only_matching': True,
  56     }, {
  57         'url': 'https://vine.co/v/bxVjBbZlPUH',
  58         'only_matching': True,
  59     }, {
  60         'url': 'https://vine.co/oembed/MYxVapFvz2z.json',
  61         'only_matching': True,
  62     }]
  63
  64     def _real_extract(self, url):
  65         video_id = self._match_id(url)
  66
  67         data = self._download_json(
  68             'https://archive.vine.co/posts/%s.json' % video_id, video_id)
  69
  70         def video_url(kind):
  71             for url_suffix in ('Url', 'URL'):
  72                 format_url = data.get('video%s%s' % (kind, url_suffix))
  73                 if format_url:
  74                     return format_url
  75
  76         formats = []
  77         for quality, format_id in enumerate(('low', '', 'dash')):
  78             format_url = video_url(format_id.capitalize())
  79             if not format_url:
  80                 continue
  81             # DASH link returns plain mp4
  82             if format_id == 'dash' and determine_ext(format_url) == 'mpd':
  83                 formats.extend(self._extract_mpd_formats(
  84                     format_url, video_id, mpd_id='dash', fatal=False))
  85             else:
  86                 formats.append({
  87                     'url': format_url,
  88                     'format_id': format_id or 'standard',
  89                     'quality': quality,
  90                 })
  91         self._check_formats(formats, video_id)
  92         self._sort_formats(formats)
  93
  94         username = data.get('username')
  95
  96         alt_title = format_field(username, template='Vine by %s')
  97
  98         return {
  99             'id': video_id,
 100             'title': data.get('description') or alt_title or 'Vine video',
 101             'alt_title': alt_title,
 102             'thumbnail': data.get('thumbnailUrl'),
 103             'timestamp': unified_timestamp(data.get('created')),
 104             'uploader': username,
 105             'uploader_id': data.get('userIdStr'),
 106             'view_count': int_or_none(data.get('loops')),
 107             'like_count': int_or_none(data.get('likes')),
 108             'comment_count': int_or_none(data.get('comments')),
 109             'repost_count': int_or_none(data.get('reposts')),
 110             'formats': formats,
 111         }
 112
 113
 114 class VineUserIE(InfoExtractor):
 115     IE_NAME = 'vine:user'
 116     _VALID_URL = r'https?://vine\.co/(?P<u>u/)?(?P<user>[^/]+)'
 117     _VINE_BASE_URL = 'https://vine.co/'
 118     _TESTS = [{
 119         'url': 'https://vine.co/itsruthb',
 120         'info_dict': {
 121             'id': 'itsruthb',
 122             'title': 'Ruth B',
 123             'description': '| Instagram/Twitter: itsruthb | still a lost boy from neverland',
 124         },
 125         'playlist_mincount': 611,
 126     }, {
 127         'url': 'https://vine.co/u/942914934646415360',
 128         'only_matching': True,
 129     }]
 130
 131     @classmethod
 132     def suitable(cls, url):
 133         return False if VineIE.suitable(url) else super(VineUserIE, cls).suitable(url)
 134
 135     def _real_extract(self, url):
 136         mobj = self._match_valid_url(url)
 137         user = mobj.group('user')
 138         u = mobj.group('u')
 139
 140         profile_url = '%sapi/users/profiles/%s%s' % (
 141             self._VINE_BASE_URL, 'vanity/' if not u else '', user)
 142         profile_data = self._download_json(
 143             profile_url, user, note='Downloading user profile data')
 144
 145         data = profile_data['data']
 146         user_id = data.get('userId') or data['userIdStr']
 147         profile = self._download_json(
 148             'https://archive.vine.co/profiles/%s.json' % user_id, user_id)
 149         entries = [
 150             self.url_result(
 151                 'https://vine.co/v/%s' % post_id, ie='Vine', video_id=post_id)
 152             for post_id in profile['posts']
 153             if post_id and isinstance(post_id, compat_str)]
 154         return self.playlist_result(
 155             entries, user, profile.get('username'), profile.get('description'))